Fluid-Structure Interaction Modeling of Pulmonary Artery Blood Flow in End-Stage Renal Disease Patients
If unaddressed, end-stage renal disease (ESRD) can result in impaired renal function, development of cardiovascular diseases, and death. Although effective, dialysis conducted via arteriovenous fistulas (AVF) has been linked to the development of pulmonary hypertension (PH). The objective of this work was to use magnetic resonance imaging (MRI) and computational fluid dynamics (CFD) to model pulmonary artery (PA) blood flow and PA fluid-structure interactions so properties of these models could be related to clinical parameters relevant to PH and ESRD. Due to the global pandemic, one (n =1) adult ESRD patient underwent cardiac MRI for PA reconstruction. Short-axis cine scans and phase-contrast scans were used respectively to recreate the patient's PA geometry in Mimics 20.0, an imaging segmentation software, and derive a patient-specific velocity waveform of blood flow at the PA inlet. Computational fluid dynamics (CFD) and finite element models of vessel deformation were created in ANSYS Workbench 19.1. Steady-state models (CFD, linear elastic fluid-structure interaction (FSI), and hyperelastic FSI) were run using a constant inlet velocity and, where applicable, mechanical properties for vessel walls taken from literature. Transient models (CFD and linear elastic FSI) were also created to simulate pulsatile blood flow and, where applicable, the interaction with the artery wall. Moderate differences were observed among the various steady-state models, whereas only minimal differences were found between transient models. These differences may be due to limited mesh resolution, which was later identified in a mesh sensitivity analysis. Although only minor differences were found between fluid results for the transient CFD and FSI model, the computational cost of the FSI model was much greater. FSI models of PA hemodynamics may not offer different results for hemodynamic properties, but they may be used in future studies when arterial remodeling and cellular stress and strain distributions are of interest. This study offered insight into computational modeling of PA blood flow and may be used as reference for future studies seeking to apply this methodology to a larger patient population. Eventually, this type of modeling may be used to identify metrics of interest that physicians can use to more closely monitor AVF development during hemodialysis.
East Carolina University